Skip to main content
Fig. 6 | BMC Bioinformatics

Fig. 6

From: SeqPredNN: a neural network that generates protein sequences that fold into specified tertiary structures

Fig. 6

Divergence between AlphaFold models of SeqPredNN predicted sequences and the native crystal structure. Sequences with high residual-TM values for the AlphaFold model of the native sequence are indicated by red crosses. a Scatter plot of the RMSD for the AlphaFold models of 662 SeqPredNN compared to the RMSD of the AlphaFold models of the native sequence. b Scatter plot of the residual-TM for the AlphaFold models of 662 SeqPredNN compared to the residual-TM of the AlphaFold models of the native sequence. The regression line \({\text{residual-TM}}_{\text{SeqPredNN/AF}} =0.809\cdot {\text{residual-TM}}_{\text{Native}/\text{AF}}+0.335\) is shown with the standard error in the shaded area. c Scatter plot of the estimated divergence of proteins generated by SeqPredNN from the native structure (\({\text{residual-TM}}_{{{\text{SeqPredNN}}/{\text{AF}}}}{-}{\text{residual-TM}}_{{{\text{Native}}/{\text{AF}}}} )\). The linear model \(-0.191\cdot {\text{residual-TM}}_{\text{Native}/\text{AF}}+0.335\) is shaded with the standard error. d Scatter plot of the estimated SeqPredNN error against the proportion of residues predicted correctly in the sequencs. A least-squares line of best fit is shown shaded with the standard error

Back to article page